pKa determination by 1H NMR spectroscopy – An old methodology revisited

Journal of Pharmaceutical and Biomedical Analysis 93 (2014) 147–155

Contents lists available at ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba

pKa determination by 1 H NMR spectroscopy – An old methodology revisited Jacqueline Bezenc¸on a,1 , Matthias B. Wittwer a,1 , Brian Cutting a , Martin Smieˇsko a , Bjoern Wagner b , Manfred Kansy b , Beat Ernst a,∗ a b

Institute of Molecular Pharmacy, Pharmacenter, University of Basel, Klingelbergstrasse 50, CH-4056 Basel, Switzerland Pharmaceuticals Division, Non-Clinical Safety, F. Hoffmann-La Roche Ltd., 4070 Basel, Switzerland

a r t i c l e

i n f o

Article history: Received 20 September 2013 Received in revised form 12 December 2013 Accepted 16 December 2013 Available online 24 December 2013 Keywords: pKa determination 1 H NMR spectroscopy Physicochemical properties Dissociation constant Site of protonation

a b s t r a c t pKa values of acids and protonated bases have an essential impact on organic synthesis, medicinal chemistry, and material and food sciences. In drug discovery and development, they are of utmost importance for the prediction of pharmacokinetic and pharmacodynamic properties. To date, various methods for the determination of pKa values are available, including UV-spectroscopic, potentiometric, and capillary electrophoretic techniques. An additional option is provided by nuclear magnetic resonance (NMR) spectroscopy. The underlying principle is the alteration of chemical shifts of NMR-active nuclei (e.g., 13 C and 1 H) depending on the protonation state of adjacent acidic or basic sites. When these chemical shifts are plotted against the pH, the inflection point of the resulting sigmoidal curve defines the pKa value. Although pKa determinations by 1 H NMR spectroscopy are reported for numerous cases, the potential of this approach is not yet fully evaluated. We therefore revisited this method with a diverse set of test compounds covering a broad range of pKa values (pKa 0.9–13.8) and made a comparison with four commonly used approaches. The methodology revealed excellent correlations (R2 = 0.99 and 0.97) with electropotentiometric and UV spectroscopic methods. Moreover, the comparison with in silico results (Epik and Marvin) also showed high correlations (R2 = 0.92 and 0.94), further confirming the reliability and utility of this approach. © 2014 Elsevier B.V. All rights reserved.

1. Introduction pKa values of acids and protonated bases have an essential impact on organic synthesis, medicinal chemistry, and material and food sciences. In drug discovery and development, they are of utmost importance for the prediction of pharmacokinetic and pharmacodynamic properties such as permeation through biological barriers, interactions with targets, or induction of side effects [1]. It is therefore indispensable to have access to this parameter at an early stage of a medicinal chemistry program. For the determination of pKa values, various experimental approaches are commonly used. First, protonation and deprotonation can be followed by UV-spectroscopy [2]. Second potentiometric measurements, also called the pH-metric approach, enable the determination of pKa values [3]. Third, with capillary electrophoresis, an additional method is available. As the migration time of a compound through a capillary depends on the fraction of charged compound, the pKa value is derived from the inflection

∗ Corresponding author. Tel.: +41 61 267 1551; fax: +41 61 267 1552. E-mail address: [email protected] (B. Ernst). 1 These authors contributed equally to this work. 0731-7085/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jpba.2013.12.014

point of the sigmoidal curve obtained when the retention times are plotted against the pH of various buffers [4]. Finally, in silico pKa prediction offers an alternative to experimental methods [5]. NMR spectroscopy offers a valuable alternative, since the protonation state of adjacent centers relates to the chemical shifts of NMR-active nuclei [6] and an NMR spectrometer is available in most chemists’ laboratories. Although specific approaches based on 15 N, 11 B, 19 F or 31 P NMR spectroscopy are reported [7–10], the pHdependence of chemical shifts of 13 C nuclei adjacent to ionizable groups is a more general approach [11] and has successfully been employed for the determination of pKa values of both acids [6,12] and protonated bases [13]. Additionally, 13 C NMR spectroscopy (often in combination with 1 H or 15 N NMR) allows the determination of pKa values in isotopically enriched proteins. The applicability has been demonstrated with, e.g., calmodulin [14]. Furthermore, pH-implications on structure (see e.g. [15]) and functionality (see e.g. [16]) of peptides and proteins can be investigated as well. If no isotope enrichment is available, the low natural abundance of 13 C nuclei (1.11%) results in prolonged recording times, representing a substantial drawback. In contrast, 1 H nuclei have a higher natural abundance, which leads to a sensitivity improvement by a factor of 5600 compared to 13 C, enabling the use of smaller amounts of sample as well

148

J. Bezenc¸on et al. / Journal of Pharmaceutical and Biomedical Analysis 93 (2014) 147–155

Fig. 1. 1 H chemical shift dependence on the ionization state of the analyte paracetamol. When plotted against the corresponding pH values, the pKa value corresponds to the inflection point of the resulting sigmoidal curve. The downfield doublet (resonance 1) corresponds to the two protons in meta-position to the hydroxy-group, whereas the upfield resonance 2 represents those at the ortho-position. The pKa derived from both resonances is 9.80 after correction for D2 O, compared to 9.63 determined by an electropotentiometric measurement [22].

as shorter recording times. The first applications were published in 1973 using a 60 MHz continuous wave spectrometer [17], and an adapted procedure to FT-NMR spectrometers was published in 2012 [18]. Moreover, an automated NMR controlled titration was described in 1995 [19]. The approach has found broad application in biology, i.e., for the pKa determination of amino acids [20], or whole proteins such as human thioredoxin [21]. In summary, despite its occasional use the broad applicability of 1 H NMR for pKa determinations in drug discovery has not been fully demonstrated. Furthermore, the method has not been extensively validated against other approaches. We therefore explored its scope and limitation and compared the results with two experimental methods, namely potentiometric [22] and UVspectroscopic (measured by Sirius SGA, spectral gradient analysis, according to [23]) pKa determinations. In addition, a comparison with two in silico tools (Epik [24], based on the Hammett and Taft empirical equations and Marvin [25], based on the calculation of partial charge of atoms) completed the validation process.

1,10-phenanthroline and 2,2 -bipyridine were purchased from Riedel-de Haën. Codeine phosphate, acetylsalicylic acid, benzocaine, and phenazone were purchased from Siegfried AG. 4-aminopyridine-2-carboxylic acid was purchased from Apollo Scientific. 2-amino-1,10-phenanthroline was purchased from Specs. 3 -(␣-d-mannopyranosyloxy)-biphenyl-4-carboxylate and 3 -chloro-4 -(␣-d-mannopyranosyloxy)-biphenyl-4-carboxylate were synthesized in house [26]. 2-Naphthacenecarboxamide was obtained from F. Hoffmann-La Roche (Basel, Switzerland). 2.2. Principle of pKa determination by 1 H NMR Chemical shifts of NMR-active nuclei (e.g., 13 C and 1 H) depend on the chemical environment, including the protonation state of adjacent acidic or basic sites. Therefore, gradual alteration of the pH causes changes in the chemical shifts that can be plotted against the pH. The pKa of the corresponding protonation site can then be extracted from the inflection point of the resulting sigmoidal curves (see Fig. 1).

2. Materials and methods 2.3. pKa determination by 1 H NMR 2.1. Instrument and chemicals NMR experiments were performed on a Bruker Avance III 500 MHz spectrometer equipped with a 5 mm PABBO probe with Z gradients at a temperature of 298 K. Spectra were recorded and processed with Topspin 2.1 (Bruker, Switzerland). Chemical shifts are given in ppm. 1D 1 H NMR experiments were performed with presaturation in order to suppress the residual HDO signal. Deuterium oxide and deuterated d6-dimethyl sulfoxide were purchased from Armar Chemicals (Switzerland). Dimethyl sulfoxide, N-acetylneuraminic acid, diazepam, glycine, l-arginine, l-cysteine, l-lysine, phenol, papaverine, propranolol, paracetamol, neocuproine, 5-nitro-1,10-phenanthroline, 4,7-dimethoxy-1,10phenanthroline, 4-phenylbutylamine, serotonin, benzoic acid, aniline, phenobarbital, diltiazem, promethazine, sucrose, 2,6dimethylaniline, 2,6-dimethylphenol, 2,2-dichloropropionic acid and uracil were purchased from Sigma. Imidazole was purchased from Acros Organics. Dioxane was obtained from Fluka.

A 100 ␮M solution of the analyte in D2 O was prepared and dioxane (50 ␮M) was added as an internal standard (at 3.75 ppm [27]). For compounds which were pre-dissolved in DMSO, DMSO was taken as the internal standard (at 2.71 ppm [27]). The solution of 100 ␮M was aliquoted into 9–15 samples and the pH values of the samples were adjusted with 0.5 M or 8 M NaOH and 0.5 M or 4 M HCl, respectively. The pH values were chosen close to the predicted pKa (software CS ChemBioDraw® Ultra, version 11.0.1), at the pKa itself, and at two extreme pH values, e.g. 1 and 13.5. Aliquots (500 ␮l) of samples at each pH value were transferred to 5 mm NMR tubes (Armar Chemicals, Switzerland) and the recorded spectra were analyzed with MestReNova (Mestrelab research, version 5.2.5-4119). The chemical shifts of the reporter proton close to the ionizable center were plotted against the pH of the solution. The resulting sigmoidal curve was subjected to nonlinear curve fitting using Prism® (GraphPad, version 5.0b and 5.0f). The pKa was obtained as the inflection point of the fitted curve. Since the

J. Bezenc¸on et al. / Journal of Pharmaceutical and Biomedical Analysis 93 (2014) 147–155

149

measurements were performed in D2 O, 0.45 pH units were subtracted to yield the corrected pKa values for a nondeuterated environment [28]. When sample consumption is critical, the titration is done progressional in a single NMR sample, lowering the compound consumption to as little as 25 ␮g for an analyte with a molecular weight of 500 g/mol. Furthermore, when small volume NMR tubes (200 ␮L compared to 500 ␮L of a standard tube of 5 mm diameter) are used, compound consumption can be further lowered to 10 ␮g. As a consequence, the pKa value of compounds with a solubility of as little as 50 ␮g/mL can still be analyzed. Lower solubility can be improved by using different solvents (e.g., CD3 OD, CD3 CN), allowing the determination of pKa values of sparsely soluble compounds. When water/organic solvent mixtures are used, the influence of the organic solvent on the pKa value can be corrected by applying Yasuda–Shedlovsky extrapolations [29,30]. Furthermore, since NMR is a non-destructive method, the analyte can be recovered provided that no pH-dependent decomposition occurred. The overall recording time is typically in the range of less than 5 min per data point and thereby reduced by a factor of 10 compared to 13 C NMR spectroscopy. A further advantage of the 1 H NMR method is the possibility to start from a DMSO stock solution as usually prepared for biological testing. In contrast to methods in earlier publications (see e.g. [28]) HCl and NaOH were used instead of DCl and NaOD for the pH-adjustment of the samples. The small amounts of added base or acid did not generate spectral disturbances when the H2 O signal was suppressed (see e.g. [20]).

3. Results and discussion For the determination of pKa values various methods are available, namely UV-spectroscopic [2], potentiometric [3], and capillary electrophoretic [4] techniques. These experimental methods offer the advantage of being highly compatible with automation, and therefore, are promising candidates for medium- to highthroughput applications. Moreover, overlapping pKa values for compounds can be determined and, with an appropriate co-solvent, pKa values for compounds with low solubility can be determined as well. Major drawbacks of these three approaches for pKa determination are the inaccessibility of information on the sites of protonation and the requirement of expensive, automated instrumentation, or laborious and time-consuming manual titration experiments. In addition, the required amount of sample is often critical. Typically, only milligram amounts of a test compound are available and the investment in expensive equipment is not always justified for single pKa determinations. Furthermore, sample impurities can lead to distorted results, with the exception of capillary electrophoresis due to its separation technique. Additionally, the UV-spectroscopic approach relies on chromophores located in proximity to the ionizable centers. In contrast to the experimental approach, in silico pKa predictions are fast and inexpensive. Nevertheless, commercial software needs to be trained and validated when applied to a new class of compounds [5]. For the evaluation of the 1 H NMR approach a diverse set of test compounds covering a broad range of pKa values (pKa 0.9–13.8) was selected (Table 1). Furthermore, pKa values were measured for compounds with up to three ionizable centers. The excellent correlation of pKa values over a range of 12 orders of magnitude with reference values determined by electropotentiometry [22] and UVspectroscopy (measured by Sirius SGA according to [23]) is reflected by R2 values of 0.99 and 0.97, respectively (Fig. 2A). In addition, the absolute average deviation of 1 H NMR pKa values compared to UV-spectroscopic and EPD were ±0.37 and ±0.25, respectively.

Fig. 2. (A) Comparison of published data from electropotentiometry [22] (black dots) and data obtained from UV-spectroscopic (orange triangles) measurements with values obtained by the presented 1 H NMR method; (B) comparison of 1 H NMR determined pKa values with computed values calculated with Epik [24] (black dots) and Marvin [25] (orange triangle) (cf. Table 1). pKa values determined by 1 H NMR are indicated as an average of the inflection points deduced from the detectable chemical shifts.

Moreover, the comparison with in silico results also exhibits a high 2 2 correlation (REpik = 0.92, RMarvin = 0.94) (Fig. 2B). A major advantage of this approach is the accessibility of pKa s of different sites of protonation, since the chemical shift differences of the reporter protons depend on the distance from the ionizable center. For example, for serotonin (Table 1, entry 34, Fig. 3B and C) two pKa values can be allocated (pKa 1 : 10.02 ±0.03 for the NH2 group, based on the chemical shifts of the protons H8 and H9 and pKa 2 : 10.89 ±0.09 for the phenolic OH, based on the chemical shifts of the protons H2 , H4 , H6 , and H7 ). As demonstrated with the examples of codeine (Table 1, entry 22), paracetamol (entry 26 and Fig. 1),

150

J. Bezenc¸on et al. / Journal of Pharmaceutical and Biomedical Analysis 93 (2014) 147–155

Table 1 pKa values obtained by 1 H NMR spectroscopy compared to published data obtained by electropotentiometric determination (EPD) (values are taken from one source [22], unless indicated otherwise); values from UV-spectroscopic determinations were measured with a Sirius SGA according to [23]. Furthermore, in silico data (Epik [24] and Marvin [25]) are shown. n.d. = not determined. Entry 2 was measured by potentiometric titrations according to [31] using a GLpKa from Sirius Analytical Instruments Ltd. Entry

Compound name and structure (ionizable centers in red; reporter protons in bold)

Conc. [mM]

Ionizable group

pKa determined by 1 H chemical shifts (reporter proton)

pKa by UV spectroscopy

pKa by EPD [22]

In silico pKa

Epik [24]

Marvin [25]

Compounds with a single pKa

0.1

H NPh

0.91 (CH3 )

n.d.

1.44

2.53

0.37

0.1

COOH

1.05 (CH3 )

–

1.87

1.25

2.63

0.1

NH3

2.18 (o-H) 2.17 (m-H) 2.16 (H1 ) 2.17 (CH3 )

n.d.

2.39

2.03

2.78

4

1

COOH

2.45 (H7 )

n.d.

2.60

2.85

2.72

5

0.1

H NCH3

2.98 (H9 )

n.d.

3.35

5.10

2.92

6

0.1

N H

3.21 (H3,8 )

3.66

3.57 [32]

3.88

2.11

7

0.1

COOH

3.20 (o-H)

n.d.

3.50

4.57

3.41

8

0.1

COOH

3.56 (H2,6 )

3.64

–

4.21

4.06

9

0.1

NH3

3.60 (m-H) 3.65(p-H) 3.60 (CH3 )

4.14

–

3.53

4.31

1

O Cl Cl

2

OH CH3

2,2-dichloropropionic acid

H

O

H

3

H1 H1 O

H2 N

H

CH3

H benzocaine

J. Bezenc¸on et al. / Journal of Pharmaceutical and Biomedical Analysis 93 (2014) 147–155

151

Table 1 (Continued) Entry

Compound name and structure (ionizable centers in red; reporter protons in bold)

Conc. [mM]

Ionizable group

pKa determined by 1 H chemical shifts (reporter proton)

pKa by UV spectroscopy

pKa by EPD [22]

In silico pKa

Epik [24]

Marvin [25]

10

0.1

COOH

4.05 (H3,5 )

3.62

–

4.29

4.07

11

0.1

COOH

3.97 (o-H) 3.96 (m-H) 3.96 (p-H)

n.d.

3.98

4.20

4.08

12

0.1

NH3

4.30 (o-H) 4.32 (m-H) 4.30 (p-H)

n.d.

4.60 [33]

4.10

4.64

13

0.1

N H

4.50 (H6,6 )

4.54

4.44 [34]

4.50

3.30

14

0.1

N H

4.65 (H2,9 ) 4.68 (H3,8 ) 4.69 (H4,7 ) 4.69 (H5,6 )

5.00

4.80 [35]

4.92

3.00

15

0.1

N H

5.33 (H2,9 )

5.57

5.78 [32]

5.08

3.46

16

0.1

N H

5.61 (CH3 )

5.74

6.15 [35]

5.41

3.65

17

1

N H

6.14 (H3 )

n.d.

6.39

6.67

6.03

18

0.1

6.38 (CH3 )

6.69

–

5.60

4.26

N H



152

J. Bezenc¸on et al. / Journal of Pharmaceutical and Biomedical Analysis 93 (2014) 147–155

Table 1 (Continued) Entry

Compound name and structure (ionizable centers in red; reporter protons in bold)

Conc. [mM]

Ionizable group

pKa determined by 1 H chemical shifts (reporter proton)

pKa by UV spectroscopy

pKa by EPD [22]

In silico pKa

Epik [24]

Marvin [25]

19

0.1

N H

6.40 (H5 )

6.73

–

5.52

4.33

20

0.1

N H

6.96 (H2 ) 6.96 (H4,5 )

n.d.

6.95 [36]

6.67

6.97

21

1

NH

7.23 (H1 )

n.d.

7.49

6.64

8.14

22

1

H NCH3

8.14 (N-CH3 ) 8.11 (O-CH3 )

n.d.

8.22

9.42

9.19

23

1

N (CH3)2 H

7.79 (CH3 )

n.d.

8.02

9.16

8.18

24

0.1

N (CH3)2 H

9.12 (CH3 )

n.d.

9.00

8.75

9.14

25

0.1

NH2

9.31 (H )

n.d.

9.53

9.09

9.67

26

10

OH

9.80 (o-H) 9.80 (m-H) 9.79 (CH3 )

n.d.

9.63

10.05

9.46

27

0.1

OH

9.92 (o-H) 9.92 (m-H) 9.91 (p-H)

n.d.

10.01

9.92

10.02

28

0.5

NH3

10.43 (CH3 )

n.d.

10.50

9.94

10.04

J. Bezenc¸on et al. / Journal of Pharmaceutical and Biomedical Analysis 93 (2014) 147–155

153

Table 1 (Continued) Entry

Compound name and structure (ionizable centers in red; reporter protons in bold)

Conc. [mM]

Ionizable group

pKa determined by 1 H chemical shifts (reporter proton)

pKa by UV spectroscopy

pKa by EPD [22]

In silico pKa

Epik [24]

29

0.1

OH

10.61 (m-H) 10.62 (p-H) 11.18 (CH3 )

30

0.5

OH

Compounds with two pKa values 31

0.1

5

32

0.5

33

10.29

–

10.31

10.73

12.82 (H4 )

n.d.

12.60

13.08

13.84

COOH

0.99 (H3 ) 0.93 (H5 ) 0.98 (H6 )

n.d.

–

0.78

2.25

NH3

9.52 (H3 ) 9.53 (H5 ) 9.63 (H6 )

9.43

–

9.32

9.00

COOH

2.92 (H2 )

n.d.

2.34 [37]

2.48

2.31

NH3

9.65 (H2 )

n.d.

9.60 [37]

8.75

9.24

NH1

9.34 (H5 ) 9.31 (H6 ) 13.76 (H5 ) 13.50 (H6 )

n.d.

9.21

9.10

9.77

n.d.

13.28

9.45

13.79

10.00 (H9 ) 10.04 (H8 ) 10.77 (H2 ) 10.99 (H4 ) 10.90 (H6 ) 10.91 (H7 )

n.d.

9.97

9.73

9.31

n.d.

10.91

9.84

10.00

COOH

1.63 (H4 ) 1.31 (H3 ) 1.54 (H2 )

n.d.

2.17 [37]

2.12

2.41

NH3

8.86 (H2 ) 9.34 (H4 ) 8.90 (H3 )

n.d.

9.04 [37]

8.77

9.12

NH2

12.08 (H3 ) 12.24 (H4 )

n.d.

12.48 [37]

12.28

12.41

COOH SH

1.54 (H2 ) 8.59 (H2 )

n.d. n.d.

1.71 [37] 8.33 [37]

1.75 8.64

2.35 9.05

NH3

10.64 (H2 )

n.d.

10.78 [37]

8.24

10.17

COOH

1.56 (H2 ) 1.61 (H4 )

n.d.

2.18 [37]

2.12

2.74

NH3

9.03 (H2 ) 9.14 (H4 )

n.d.

8.95 [37]

9.88

9.44

NH3

10.31 (H4 ) 10.53 (H6 )

n.d.

10.53 [37]

8.77

10.29

OH Ph-OH

2.65 (H4 ) 8.40 (CH3 )

3.13 7.40

– –

4.85 6.71

4.24 7.83

N (CH3)2 H

8.79 (N-CH3 )

9.35

–

9.38

9.00

2

NH

0.5

34

NH3 OH

Compounds with three pKa values 35

10

36

37

38

10

10

H 3C CH3 CH3 OH N H4 OH OH NH2 OH O

OH O

O

2-naphthacenecarboxamide

0.1

Marvin [25]

154

J. Bezenc¸on et al. / Journal of Pharmaceutical and Biomedical Analysis 93 (2014) 147–155

few minutes per pH value are required. However, even in cases where high-accuracy data are requested, compound requisition could be kept low by applying small volume NMR tubes (volume of 200 ␮L) and/or increasing recording times. Second, compounds can be recovered because the method is non-destructive. Third, pKa determination in D2 O is feasible when a correction factor is applied [28]. Fourth, an additional strength of this approach is that it allows the correlation of pKa values with the various protonation sites. Fifth, in cases where the chemical shifts of protons adjacent to the ionizable centers are not resolved, NMR spectroscopy with other nuclei, including 13 C, 15 N or 19 F, or 2D-methods, such as HSQC, could be applied. Finally, with the use of different solvents (e.g., CD3 OD, CD3 CN), solubility could be improved, allowing the determination of pKa values in sparingly soluble compounds. In conclusion, 1 H NMR spectroscopy allows a fast, sensitive and reliable determination of pKa values over a broad pH-range, and can be used for compounds with more than one ionizable center. Furthermore, it is applicable to compounds with low solubility, and only microgram amounts of an analyte are required. Finally, the results from pKa determination by 1 H NMR correlate well with data obtained with the commonly used potentiometric and UVspectroscopic methods. Acknowledgements The authors gratefully acknowledge the financial support by the Swiss National Science Foundation (Grant No. 200020120628). The authors express their gratitude to Dr. Rachel Hevey, Institute of Molecular Pharmacy, University of Basel for critical revision of the manuscript. References

Fig. 3. Chemical shifts plotted against pH. The pKa can be deduced from the inflection point of the sigmoidal curves for (A) glycine and (B,C) serotonin.

and 2,6-dimethylphenol (entry 29), equal pKa values are obtained independent of the distance of the reporter proton to the ionizable center. Glycine is an example where the pKa values of two different ionizable centers can be determined with one reporter proton (Table 1, entry 32 and Fig. 3A). Moreover, the applicability to compounds with more than two ionizable centers is demonstrated in entries 35–38. However, not every chemical shift of the 1 H NMR spectra can be used for the analysis because the resonance of the internal standard (e.g. DMSO or dioxane) or of H2 O may overlap with resonances of reporter protons. 4. Conclusions For the determination of pKa values by 1 H NMR, several features need to be considered. First, especially in cases where only a rough estimate of pKa values (±1.00) rather than high-accuracy data is required, the methodology is highly valuable because compound requisition in the microgram range and recording times of only a

[1] D. Horter, J.B. Dressman, Influence of physicochemical properties on dissolution of drugs in the gastrointestinal tract, Adv. Drug Deliv. Rev. 46 (2001) 75–87. [2] K. Box, C. Bevan, J. Comer, A. Hill, R. Allen, D. Reynolds, High-throughput measurement of pKa values in a mixed-buffer linear pH gradient system, Anal. Chem. 75 (2003) 883–892. [3] A. Avdeef, pH-metric log P. II: Refinement of partition coefficients and ionization constants of multiprotic substances, J. Pharm. Sci. 82 (1993) 183–190. [4] Y. Ishihama, M. Nakamura, T. Miwa, T. Kajima, N. Asakawa, A rapid method for pKa determination of drugs using pressure-assisted capillary electrophoresis with photodiode array detection in drug discovery, J. Pharm. Sci. 91 (2002) 933–942. [5] A.C. Lee, G.M. Crippen, Predicting pKa , J. Chem. Inf. Model. 49 (2009) 2013–2033. [6] D. Farcasiu, A. Ghenciu, Determination of acidity functions and acid strengths by C-13 NMR, Prog. Nucl. Magn. Reson. Spectrosc. 29 (1996) 129–168. [7] J.W. Paschal, D.E. Dorman, Determination of pKa values using 15 N- and 13 Cnuclear magnetic resonance spectroscopy – a case of apramycin, Org. Magn. Res. 11 (1978) 632–634. [8] J.A. Tossell, Boric acid, carbonic acid, and N-containing oxyacids in aqueous solution: ab initio studies of structure, pKa , NMR shifts, and isotopic fractionations, Geochim. Cosmochim. Acta 69 (2005) 5647–5658. [9] M.C. Malet-Martino, J. Bernadou, R. Martino, J.P. Armand, 19 F NMR spectrometry evidence for bile acid conjugates of ␣-fluoro-␤-alanine as the main biliary metabolites of antineoplastic fluoropyrimidines in humans, Drug Metab. Dispos. 16 (1988) 78–84. [10] K.K. Millis, M.E. Colvin, E.M. Shulman-Roskes, S.M. Ludeman, O.M. Colvin, M.P. Gamcsik, Comparison of the protonation of isophosphoramide mustard and phosphoramide mustard, J. Med. Chem. 38 (1995) 2166–2175. [11] S. Berger, S. Braun, 200 and More NMR Experiments: A Practical Course, WileyVCH, Weinheim, 2004, pp. 290–292. [12] D.P. Cistola, D.M. Small, J.A. Hamilton, Ionization behavior of aqueous shortchain carboxylic acids – a 13 C-NMR study, J. Lipid Res. 23 (1982) 795–799. [13] M. Delfini, A.L. Segre, F. Conti, R. Barbucci, V. Barone, P. Ferruti, On the mechanism of protonation of triamines, J. Chem. Soc. Perkin Trans. 2 (1980) 900–903. [14] M.J. Zhang, H.J. Vogel, Determination of the side-chain pKa values of the lysineresidues of calmodulin, J. Biol. Chem. 268 (1993) 22420–22428. [15] A.T. Clark, K. Smith, R. Muhandiram, S.P. Edmondson, J.W. Shriver, Carboxyl pKa values, ion pairs, hydrogen bonding, and the pH-dependence of folding the hyperthermophile proteins Sac7d and Sso7d, J. Mol. Biol. 372 (2007) 992–1008. [16] E. Spink, S. Cosgrove, L. Rogers, C. Hewage, J.P.G. Malthouse, C-13 and H-1 NMR studies of ionizations and hydrogen bonding in chymotrypsin-glyoxal inhibitor complexes, J. Biol. Chem. 282 (2007) 7852–7861. [17] C.S. Handloser, M.R. Chakrabarty, M.W. Mosher, Experimental determination of pKa values by use of NMR chemical shifts, J. Chem. Educ. 50 (1973) 510–511.

J. Bezenc¸on et al. / Journal of Pharmaceutical and Biomedical Analysis 93 (2014) 147–155 [18] A.D. Gift, S.M. Stewart, P.K. Bokashanga, Experimental determination of pKa values by use of NMR chemical shifts, revisited, J. Chem. Educ. 89 (2012) 1458–1460. [19] J. Ollig, G. Hägele, NMR-controlled titrations of phosphorus-containing acids and bases, Comp. Chem. 19 (1995) 287–294. [20] D.L. Rabenstein, T.L. Sayer, Determination of microscopic acid dissociationconstants by nuclear magnetic resonance spectrometry, Anal. Chem. 48 (1976) 1141–1145. [21] J. Quijada, G. Lopez, R. Versace, L. Ramirez, M.L. Tasayco, On the NMR analysis of pKa values in the unfolded state of proteins by extrapolation to zero denaturant, Biophys. Chem. 129 (2007) 242–250. [22] A. Avdeef, Absorption and Drug Development: Solubility, Permeability, and Charge State, Wiley-Interscience, Hoboken, NJ, 2003. [23] F. Milletti, L. Storchi, L. Goracci, S. Bendels, B. Wagner, M. Kansy, G. Cruciani, Extending pKa prediction accuracy: high-throughput pKa measurements to understand pKa modulation of new chemical series, Eur. J. Med. Chem. 45 (2010) 4270–4279. [24] J.C. Shelley, A. Cholleti, L.L. Frye, J.R. Greenwood, M.R. Timlin, M. Uchimaya, Epik: a software program for pKa prediction and protonation state generation for drug-like molecules, J. Comput. Aided Mol. Des. 21 (2007) 681–691. [25] Marvin 5.6.0.2, ChemAxon, 2011, http://www.chemaxon.com/marvin/help/ calculations/pKa.html (23.11.13). [26] T. Klein, D. Abgottspon, M. Wittwer, S. Rabbani, J. Herold, X. Jiang, S. Kleeb, C. Lüthi, M. Scharenberg, J. Bezenc¸on, E. Gubler, L. Pang, M. Smieˇsko, B. Cutting, O. Schwardt, B. Ernst, FimH antagonists for the oral treatment of urinary tract infections: from design and synthesis to in vitro and in vivo evaluation, J. Med. Chem. 53 (2010) 8627–8641. [27] H.E. Gottlieb, V. Kotlyar, A. Nudelman, NMR chemical shifts of common laboratory solvents as trace impurities, J. Org. Chem. 62 (1997) 7512–7515. [28] A. Krezel, W. Bal, A formula for correlating pKa values determined in D2 O and H2 O, J. Inorg. Biochem. 98 (2004) 161–166.

155

[29] K. Sarmini, E. Kenndler, Ionization constants of weak acids and bases in organic solvents, J. Biochem. Biophys. Methods 38 (1999) 123–137. [30] A. Avdeef, K.J. Box, J.E.A. Comer, M. Gilges, M. Hadley, C. Hibbert, W. Patterson, K.Y. Tam, pH-metric log P. 11: pKa determination of water-insoluble drugs in organic solvent-water mixtures, J. Pharm. Biomed. Anal. 20 (1999) 631–641. [31] J. Olsen, P. Seiler, B. Wagner, H. Fischer, T. Tschopp, U. Obst Sander, D.W. Banner, M. Kansy, K. Müller, F. Diederich, A fluorine scan of the phenylamidinium needle of tricyclic thrombin inhibitors: effects of fluorine substitution on pKa and binding affinity and evidence for intermolecular C F. . .CN interactions, Org. Biomol. Chem. 2 (2004) 1339–1352. [32] T. Ramirez-Silva, M. Gomez-Hernandez, D.L. Pacheco-Hernandez, A. RojasHernandez, L. Galicia, Spectroscopy study of 5-amino-1,10-phenanthroline, Spectrochim. Acta A 60 (2004) 781–789. [33] Y. Altun, Study of solvent composition effects on the protonation equilibria of various anilines by multiple linear regression and factor analysis applied to the correlation between protonation constants and solvatochromic parameters in ethanol–water mixed solvents, J. Solution Chem. 33 (2004) 479–497. [34] M. Yasuda, K. Sone, K. Yamasaki, Stability of zinc and cadmium complexes with some methyl derivatives of 1,10-phenantroline and 2,2 -bipyridine, J. Phys. Chem. 60 (1956) 1667–1668. [35] W.W. Brandt, D.K. Gullstrom, Studies on some ferrous complexes of substituted 1,10-phenantrolines, J. Am. Chem. Soc. 74 (1952) 3532–3535. [36] T.C. Bruice, G.L. Schmir, Imidazole catalysis. 2. The reaction of substituted imidazoles with phenyl acetates in aqueous solution, J. Am. Chem. Soc. 80 (1958) 148–156. [37] R.M.C. Dawson, D.C. Elliott, W.H. Elliott, K.M. Jones, Data for Biochemical Research, Clarendon Press, Oxford, 1959.